Addition of Estradiol to Cross-Sex Testosterone Therapy Reduces Atherosclerosis Plaque Formation in Female ApoE^−/− Mice

Abstract

The contributions of estradiol and testosterone to atherosclerotic lesion progression are not entirely understood. Cross-sex hormone therapy (XHT) for transgender individuals dramatically alters estrogen and testosterone levels and consequently could have widespread consequences for cardiovascular health. Yet, no preclinical research has assessed atherosclerosis risk after XHT. We examined the effects of testosterone XHT after ovariectomy on atherosclerosis plaque formation in female mice and evaluated whether adding low-dose estradiol to cross-sex testosterone treatments after ovariectomy reduced lesion formation. Six-week-old female ApoE^−/− C57BL/6 mice underwent ovariectomy and began treatments with testosterone, estradiol, testosterone with low-dose estradiol, or vehicle alone until euthanized at 23 weeks of age. Atherosclerosis lesion progression was measured by Oil Red O stain and confirmed histologically. We found reduced atherosclerosis in the estradiol- and combined testosterone/estradiol–treated mice compared with those treated with testosterone or vehicle only in the whole aorta (−75%), aortic arch (−80%), and thoracic aorta (−80%). Plaque size was similarly reduced in the aortic sinus. These reductions in lesion size after combined testosterone/estradiol treatment were comparable to those obtained with estrogen alone. Testosterone/estradiol combined therapy resulted in less atherosclerosis plaque formation than either vehicle or testosterone alone after ovariectomy. Testosterone/estradiol therapy was comparable to estradiol replacement alone, whereas mice treated with testosterone only fared no better than untreated controls after ovariectomy. Adding low-dose estrogen to cross-sex testosterone therapy after oophorectomy could improve cardiovascular outcomes for transgender patients. Additionally, these results contribute to understanding of the effects of estrogen and testosterone on atherosclerosis progression.

Adding low-dose estradiol to cross-sex testosterone therapy in female mice after ovariectomy improved atherosclerosis plaque formation by 75% to 80%, comparable to effects of estradiol replacement alone.

The sex steroid hormones estradiol (E) and testosterone (T) are known to influence atherosclerosis formation and progression both in humans and experimental animals. The effect of these hormones is confounded by sex. Perhaps the most dramatic example allowing uncoupling of the effects of sex and sex steroid hormones is in the transgender population. Transgender individuals identify as a gender that differs from the one they were assigned at birth and make up an estimated 1,400,000 people in the United States, ∼0.6% of the national population (1). These individuals frequently undergo cross-sex hormone therapy (XHT) to alter secondary sex characteristics. For female-to-male (FtM) transgender patients, this involves lowering endogenous ovarian hormone levels (often by oophorectomy) and supplementing exogenous T to levels typical of nontransgender (cisgender) males. This can induce body and facial hair growth, voice deepening, and altered fat distribution (2). Although widespread statistics are not available, a recent study documented that 97% of transgender patients treated in a clinic were using XHT (3). XHT is generally used continuously and indefinitely, yet the long-term implications are unknown. Because gonadal steroids are involved in numerous physiological processes, XHT would be expected to cause changes that are not limited to the reproductive tract and secondary sex characteristics. Although limited research has been executed on long-term effects of XHT, observational studies have suggested various adverse health outcomes (4–7). Yet, those limited studies lacked both controls and long-term follow up (particularly for FtM individuals), suggesting the need for basic and translational research.

Atherosclerosis is the most common cause of death in Western societies. This cardiovascular disease is a chronic inflammatory process involving interactions among lipoproteins, monocyte-derived macrophages, T cells, and the arterial wall, which results in development of plaques or lesions that project out into the lumen of the vessel (8). Cardiovascular disease displays marked sex differences (9), with men twice as likely to die of cardiovascular disease as women (10, 11), and a growing body of research suggesting sex differences in the cells involved in human atherosclerosis (12). Because gonadal steroid hormones are implicated in cardiac health (11, 13), atherosclerosis risk is of particular concern for patients on XHT, although effects have not been previously studied.

Estrogens have been proposed to reduce atherosclerosis (11, 14–19), whereas T does not appear to have similar protective functions, yet data on this facet of endocrinology remain controversial. The 1998 Heart and Estrogen/Progestin Replacement Study (20) found no benefit to postmenopausal estrogen therapy in women with preexisting coronary artery disease, and the 2002 Women’s Health Initiative study (21) suggested potential adverse health outcomes, including increased coronary heart disease, following postmenopausal estrogen therapy. Yet, longer-term follow-up on the Women’s Health Initiative work has since revealed that estrogen therapy for younger participants improved atherosclerosis outcomes (22), which was confirmed in the Early Versus Late Intervention Trial With Estradiol study (23). In contrast, T therapy in males has been associated with increased cardiovascular disease (24). In postmenopausal women, serum T levels are directly associated with atherosclerosis progression, but estrogen levels are inversely correlated with lesion development (13, 18, 25, 26), Preclinical work supports these findings because female mice treated with high-dose T have decreased serum high-density lipoprotein (HDL) levels and increased atherosclerosis development, mimicking the outcomes seen in males (27), whereas estrogens can have cardioprotective effects in female rabbits and primates after ovariectomy (OVX) (28, 29). Taken together, this suggests that XHT T treatments after oophorectomy for FtM individuals could increase atherosclerosis risk. However, despite the prevalence of XHT use, no preclinical models or controlled studies have yet investigated the effects of the physiologically relevant high-dose T used in females as XHT.

Using an ApoE-null mouse model, we consider the effects of T-based XHT after OVX on atherosclerosis development in female mice. ApoE is a vital ligand for the uptake and clearance of atherogenic lipoproteins, including very low-density lipoproteins (LDLs) and chylomicrons (30). Mice do not develop atherosclerosis under normal conditions; however, targeted mutation of the ApoE gene results in severe hypercholesterolemia and early development of atherosclerotic lesions (30–34), making this murine model ideal for studying factors affecting cardiac disease progression. Because estrogen replacement therapy has been shown to reduce atherosclerosis progression in younger postmenopausal women (17), we investigate if adding low-dose estrogen to T XHT might reduce atherosclerosis risk after gonad removal in adolescent female mice.

Materials and Methods

Animals

This article was prepared according to Animal Research: Reporting of In Vivo Experiments guidelines (35). All animal experiments were conducted using female Apoe^tm1Unc mice (n = 24) obtained from the Jackson Laboratory (Bar Harbor, ME) in accordance with an approved Yale University Animal Care Committee protocol. All animals received a chow diet ad libitum. At 6 weeks of age, four groups of mice (n = 6 per group), randomly assigned, underwent OVX and began weekly injections of T (group T), E benzoate (group E), T and E mixed group (T+E), or vehicle control (group C), continuing through 22 weeks of age. Mice were weighed before undergoing OVX at 6 weeks and again before being euthanized at 23 weeks.

Bilateral OVX was performed with meloxicam analgesic and isoflurane anesthetic. Hormone treatments followed our previously published paradigm (36), with weekly subcutaneous 100 µL sesame oil injections. Mice in the E-treated group received a previously characterized dose of E benzoate used as a model for postmenopausal hormone replacement therapy in female mice after OVX, weight-adjusted to average adult C57Bl/6 bodyweight, which was 6.4 μg/wk E benzoate per mouse (37). T and T+E groups received T in the weight-adjusted amount as used for human XHT (31 µg/wk) (38–42); the T+E group received E benzoate at a weight-adjusted estrogen dose matching levels in cisgender males (0.8 µg/wk).

All mice were euthanized at 23 weeks of age after overnight fasting. Plasma was collected by cardiac puncture and stored overnight at 4°C before analysis. Whole aorta was dissected and stored in formalin at 4°C until being subjected to Oil Red O (ORO) staining. Whole hearts were extracted and cut in half, and the top half for each mouse was embedded in Optimal Cutting Temperature (Sakura Finetechnical Co, Ltd) compound after overnight fixation in 10% formalin, and frozen in −80°C until sections were cut for histology.

Plasma analysis

Standard commercial enzyme-linked immunosorbent assay kits were used according to the manufacturer’s protocol to determine E (Calbiotech, Spring Valley, CA) and T (Immuno-Biological Laboratories, Inc., Minneapolis, MN) levels. Commercial kits were used to measure plasma triglyceride (TG; Diagnostic Chemicals, Charlottetown, PEI, Canada), and total cholesterol (TC) and HDL (ThermoFisher Scientific Inc., Waltham, MA) concentrations. Subtracting HDL from TC gave LDL. Samples were run in duplicate. Standard commercial kits were similarly used to quantify plasma levels of alanine aminotransferase, blood urea nitrogen, creatinine, bilirubin, insulin, and glucose, and performed in the Yale core facility laboratories.

ORO staining

ORO staining solution was prepared by mixing 35 mL ORO solution (0.2%, weight-to-volume ratio) in methanol with 10 mL of 1M sodium hydroxide and filtered with filter paper (43). Whole aortas were washed in 1 mL 78% methanol for 5 minutes on a tilted roller, incubated in 1 mL ORO staining solution on tilted rollers for 50 minutes, destained in 1 mL 78% methanol for 5 minutes, and then transferred to phosphate-buffered saline until they were mounted and imaged.

Adventitial fat was removed under an Olympus SZX16 microscope with fine forceps. Then aortas were cut longitudinally using microdissecting spring scissors and pinned flat onto clear-bottom Sylgard-coated glass dissecting dishes, as described (44). Images were captured using an Olympus SZX16 microscope with an SDF PLAPO 0.5 XPF objective lens connected to an Olympus Model U-LH100HG camera (100 W 19 V) and analyzed using ImageJ software with a [0,131] red threshold, determined based on ability to visualize red stain in the E and T+E groups.

Histology

Sections of the aortic sinus were cut, mounted, and stained with hematoxylin-eosin (H&E) for histopathological analysis. Photographs were taken using a Nikon Eclipse 80i histology microscope, with a 4×/0.13 ± WD 17.1 objective lens. Lesion size was measured for all mice using ImageJ and is presented as average lesion area per animal for each treatment group, normalized to average lesion area for OVX/vehicle mice. Three sections were evaluated per mouse. No additional processing was done. Black arrows were added using GraphPad Prism for Mac.

Statistical analysis

All statistical analyses were performed on GraphPad Prism for Mac. The primary outcome for this study is aortic plaque formation, quantified by ORO staining. One-tailed t tests were used to compare plasma E and T levels. Analysis of variance with post hoc multiple comparisons tests using a Tukey correction were used to determine differences between groups for weight change, quantitative plaque analyses, and plasma lipids and cholesterol. No mice were excluded from the study. All quantitative data are presented with mean ± standard error of the mean (SEM).

Results

Hormone levels and body weight changes

To validate our model, we first checked plasma E and T levels in all groups, at 23 weeks (5 days postinjection). E levels were significantly higher in E-treated mice (mean E ± SEM, 77.2 ± 34.3 pg/mL) than in controls (3.9 ± 0.8 pg/mL; P = 0.05) or T-treated mice (3.0 ± 0.8 pg/mL; P = 0.03). Further, combined therapy mice had higher E levels (29.6 ± 13.3 pg/mL) than both control (P = 0.05) and T-treated mice (P = 0.04). T levels were higher in T-treated mice (22.3 ± 2.1 ng/dL) compared with controls (17.2 ± 1.2 ng/dL; P = 0.04) or E-treated mice (16.3 ± 1.2 ng/dL; P = 0.02). The increase in T levels was not substantial in the combined therapy group (18.0 ± 2.0 ng/dL).

We found a significant relationship between treatment and change in body weight from 6 (pre-OVX) to 23 weeks (F = 8.5, P = 0.001). The T group (mean weight gain ± SEM, +6.4 ± 0.74 g) did not differ from vehicle (+4.8 ± 0.2 g), E (+7.9 ± 0.5 g), or combined therapy (+8.0 ± 0.5g) mice, but E and T+E combined therapy groups had comparably greater weight gain than vehicle-treated controls (E, P = 0.003; T+E, P = 0.003).

These changes in hormone levels and body weight were not associated with important changes in renal or liver function, as measured by blood urea nitrogen, creatinine, alanine aminotransferase, and bilirubin levels (Supplemental Table 1). Random glucose and insulin levels were also within the normal range and not significantly different between groups (Supplemental Table 1).

Plaque formation in whole aorta and aortic root

To investigate the effect of different treatments on atherosclerotic plaque formation, we used aortic ORO stain (relative staining with the quantification coloration threshold displayed in Fig. 1D). Treatment was related to percent ORO staining for whole aorta (F = 9.07, P = 0.006), aortic arch (F = 7.357, P = 0.002), and thoracic aorta (F = 7.659, P = 0.0015).

Figure 1.

Quantification of aorta ORO stain. Percent stained area is shown for (A) whole aorta, (B) aortic arch, and (C) thoracic aorta (vehicle control: n = 5; T, E, and T+E: n = 6 per group). Data are presented as mean ± SEM. There was a significant relationship between treatment and ORO staining for the whole aorta (F = 9.07, P = 0.006), aortic arch (F = 7.36, P = 0.002), and thoracic aorta (F = 7.67, P = 0.0015). Post hoc Tukey-corrected tests demonstrated that the E and T+E groups had significantly less lesion formation than mice treated with vehicle or with T alone, for whole aorta, aortic arch, and thoracic aorta. No other comparisons were significantly different. Significance compared with vehicle controls: *P = 0.03, **P = 0.02. Significance compared with T: ^#P = 0.03, ^##P = 0.02, ^###P < 0.01. (D) A representative aorta for each group with red quantification on ImageJ is shown.

Although T alone in the OVX group did not significantly alter lesion formation compared with vehicle, we found reduced stained lesion area in E only and combined therapy groups compared with mice treated with vehicle or T alone. This included 75% reduction in lipid plaque area in the whole aorta in E and T+E mice compared with vehicle controls (E, P = 0.02; T+E, P = 0.02; Fig. 1A) and 80% reduction compared with T-treated mice (E, P < 0.01; T+E, P < 0.01; Fig. 1A). Similar reductions were observed in the aortic arch, with E mice exhibiting 70% reduction in plaques compared with vehicle controls and T mice (vs. vehicle controls: P = 0.03, vs T: P = 0.03; Fig. 1B) and with combined therapy mice exhibiting 80% reduction in lesion formation compared with both C and T mice (vs. C: P = 0.02, vs T: P = 0.02; Fig. 1B). 80% reduction in plaques was apparent for both E and T/E groups in the thoracic aorta compared with both vehicle (E: P = 0.03, T+E: P = 0.04; Fig. 1C) and T (E, P < 0.01; T+E, P < 0.01; Fig. 1C) mice. The E and combination therapies yielded comparable results (P > 0.99).

We confirmed these findings using H&E staining on aortic root sections, measuring lesion size. Vehicle- or T-treated OVX mice displayed noticeable large plaque formation (purple in contrast to the red of the endothelial wall, indicated with black arrows; Fig. 2), yet E and combination therapy-treated OVX groups did not (Fig. 2). This observation was supported quantitatively with a significant relationship between average lesion area per H&E-stained aortic root sample and treatment (F = 10.92, P = 0.0002). Measurement of average lesion area revealed that E-treated mice lesion size was 5% that of the control (P < 0.01; Fig. 3); combined therapy mice were 11% (P < 0.01; Fig. 3). E and T+E combined treatment mice both exhibited lower average lesion area compared with T mice (E, P < 0.01; T+E, P < 0.01; Fig. 3), which were 83% the area of controls (P = 0.87). E and combined therapy results were comparable (P > 0.99).

Figure 2.

H&E staining of the aortic root. Immunohistochemistry images are shown for representative mouse groups treated with vehicle control (C), T, E, and combined T+E after OVX. Photographs were taken at ×4 magnification; black arrows indicate large plaque formation.

Figure 3.

Average lesion area in H&E-stained aortic root. Average lesion area is presented as fold change compared with C, with bars representing SEM. We found a significant relationship between average lesion area per H&E-stained aortic root sample and treatment (F = 10.92, P = 0.0002). Post hoc Tukey-corrected tests demonstrated a reduction in average lesion area for E-treated and T+E-treated mice, compared with both C- and T-treated mice. Significance compared with C, *P < 0.01. Significance compared with T, ^#P < 0.01.

Plasma cholesterol and triglycerides

To determine if changes to circulating cholesterol and TG levels contributed to these observed differential outcomes, we measured the plasma concentration of TC, HDL, LDL, and TG in all mice from all groups. Treatment was related to TC (F = 5.587, P = 0.0064), HDL (F = 3.478, P = 0.0363), LDL (F = 5.459, P = 0.0071), and TG (F = 3.409, P = 0.0387). E therapy (E) resulted in significantly decreased TC (−35%, P < 0.01; Table 1), HDL (−45%, P = 0.03; Table 1), and LDL (−35%, P < 0.01; Table 1) compared with vehicle, and reduced TC (−30%, P = 0.03; Table 1), LDL (−30%, P = 0.03; Table 1), and TG (−30%, P = 0.04; Table 1) compared with T, consistent with previous characterizations of ovariectomized ApoE^−/− mice treated with E (45). The reductions observed with combined therapy compared with vehicle and T-treated mice were not statistically significant (Table 1).

Table 1.

Plasma Lipoprotein and Lipid Concentrations

Treatment	TC (mg/dL)	HDL (mg/dL)	LDL (mg/dL)	TG (mg/dL)
C	254.4 ± 25.51	6.69 ± 0.86	247.7 ± 24.77	56.01 ± 9.08
T	235.4 ± 20.16	5.46 ± 0.82	229.9 ± 19.83	74.88 ± 6.05
E	164.8 ± 9.17a,b	3.66 ± 0.56c	161.1 ± 9.09a,b	50.84 ± 3.19d
T+E	201.6 ± 8.26	5.0 ± 0.31	196.5 ± 8.36	64.69 ± 4.46

Data are presented as mean ± SEM (C: n = 5; T, E, T+E: n = 6 per group). There was a significant relationship between treatment and TC (F = 5.587, P = 0.0064), HDL (F = 3.478, P = 0.0363), LDL (F = 5.459, P = 0.0071), and TG (F = 3.409, P = 0.0387), according to analysis of variance. Post hoc Tukey-corrected tests demonstrated that the E therapy had significantly lower levels than mice treated with vehicle TC, HDL, and LDL, and compared with T mice for TC, LDL, and TG. No other comparisons were significantly different.

Abbreviation: C, vehicle control.

^aSignificance compared with C: P < 0.01.

^bSignificance compared with T: P = 0.03.

^cSignificance compared with C: P = 0.03.

^dSignificance compared with T: P = 0.04.

Discussion

In this study, we assessed the effects of sex steroid treatment following OVX in female ApoE knockout mice on atherosclerosis development and whether adding a low dose of estrogen to T treatment reduces plaque formation. The use of T in female mice uncouples the effects of sex from that of sex steroid hormones. T administration also models cross-sex T therapy after oophorectomy in transgender female-to-male sex-affirmative treatment in humans. Our findings suggest that T does not reduce atherosclerosis progression, but estrogen does; yet, adding a low dose of estrogen can mimic the reductions in atherosclerosis lesions observed with estrogen replacement alone.

Female ApoE null mice have been previously shown to exhibit accelerated atherosclerosis development after OVX; however, estrogen therapy reverses this effect (14). Here we found a substantial reduction in atherosclerosis lesion formation in both females treated with estrogen after OVX and females treated with a combined therapy of T and a low dose of estrogen, compared with those treated with T alone after OVX. However, this estrogen-dependent reduction in plaque formation did not appear to be dependent on improvements in plasma cholesterol and TG. Consistent with prior work (16), estrogen reduced atherosclerosis progression, independent of circulating cholesterol levels. Here we showed that reductions in both LDL and HDL cholesterol in E-treated mice, but no important alterations to the ratio, which is acknowledged to be the more meaningful cholesterol metric (46). The changes that we observed in lipid levels matched previously published results as well (45). These findings are consistent with clinical work suggesting that changes in plasma cholesterol are responsible for relatively little of estrogen-dependent protection against atherosclerosis (47); numerous other mechanisms might be responsible for the observed benefits (48). Such mechanisms include decreased recruitment of macrophages (central to all stages of atherosclerosis pathogenesis and progression) (49), increased nitric oxide production (50, 51) (which reduces atherosclerotic progression) (52–56), and reduced aortic endothelial vascular cell adhesion molecule-1 expression (57, 58) (which recruits immune cells to atherosclerotic lesions to bolster plaque formation) (59–61). In contrast, androgens cause opposite effects to estrogens on these three mechanisms (62–64).

Although some males have been shown to have reduced atherosclerosis formation with T treatments (65), that effect depends on T aromatization to E (66, 67). Preclinical models have demonstrated that estrogen receptor function mediates prevention of early atherosclerosis in both females and males (68–70). Clinically, this is supported by advanced atherosclerosis development in males with aromatase deficiency (71–74), which can be reduced by estrogen therapy both clinically (18) and in preclinical models (29, 66, 75–77). Cisgender men have higher estrogen levels than postmenopausal cisgender women because of aromatization of T (78, 79). Yet, aromatization has been shown to be sexually dimorphic, with cisgender men aromatizing T twice as much (0.4% vs. 0.2%) as cisgender women (80). If aromatization of T XHT in FtM transgender patients were not sufficient to increase E levels, supplementing the T therapy with a low dose of E could improve long-term outcomes.

One limitation of this study is the small extent of T increase with the combined therapy. Estrogen therapy lowered both endogenous T levels and the increase in T achieved with therapy. This needs to be further investigated to ensure that adding low-dose E does not interfere with developing the desired secondary sex characteristics.

Species differences are an additional limitation of this work. Mice lack sex hormone binding globulin (81), which may have influenced our findings. We followed the established rodent models of T and E injections that were dosed to mimic human hormone therapy (including as a replacement of endogenous hormones for males) (82–89). We selected this treatment paradigm to model human XHT as closely as possible. Our data are consistent with previous work demonstrating that high doses of T in female mice are also associated with atherosclerosis progression (27). Similarly, although mice have low endogenous estrogen levels, the dose used here is based on published postmenopausal estrogen replacement paradigms in female mice (37) and has been used previously to model XHT (36). Further work is needed to validate these doses and our treatment model. Finally, although aromatase in cardiovascular tissue can convert T to E, both our findings and prior research (90, 91) suggest that this aromatization is insufficient to sustain estrogen levels at the level required for reducing atherosclerosis after OVX.

With this study, we investigated relative contributions of E and T to atherosclerosis lesion progression in ovariectomized female mice, and found that a low dose of E is required for reducing atherosclerosis development after gonadectomy. This has implications for hormone action on atherosclerosis formation, but is perhaps most important for transgender individuals currently using T XHT after oophorectomy without estrogen supplementation. Our findings suggest a potentially improved treatment of reducing atherosclerosis risk of FtM transgender patients. If confirmed in humans, these results would offer a large contribution toward improving sex-affirming treatment of transgender individuals to reduce risk of cardiovascular disease.

Abbreviations:

E

estradiol

FtM

female-to-male

HDL

high-density lipoprotein

H&E

hematoxylin and eosin

LDL

low-density lipoprotein

ORO

Oil Red O

OVX

ovariectomy

SEM

standard error of the mean

T

testosterone

TC

total cholesterol

T+E

testosterone and estradiol mixed group

TG

triglyceride

XHT

cross-sex hormone therapy.